Recent literature documents that adverse early-life conditions can have negative long-term effects on education, labor market outcomes, and mortality. However, there is scant evidence as to whether public policies such as income transfers to poor families can effectively improve the underlying conditions of the young. We propose the first analysis of the long-term effects of cash transfers (welfare) on recipient children by evaluating the effects of Mothers' Pension (MP) programs on mortality, disability, earnings and educational attainment. The results will better inform policy-makers regarding the potential for income transfers to reduce the impact of disparities in childhood circumstances on lifetime differences in socioeconomic status and health. MP programs (1911-1935) were the first government sponsored welfare program for families with dependent children in the U.S. and served as the basis for the federal program that is currently known as TANF. Among the state programs, there was considerable heterogeneity in eligibility (widowed only vs. inclusion of abandoned/divorced women), amount of monthly pension and duration, and conditions of receipt (unconditional, maternal work, or child school attendance). [We have collected information on over 80,000 accepted and rejected MP applicants in 14 states who applied between1911 and1935. We are in the process of matching these individual data with information from the Social Security Administration that includes date of death, disability application/receipt, and earnings, as well as to information from the 1940 census and World War II (WWII) enlistment records on years of completed schooling, occupation, height and IQ, and 1940 income (Aim 1). To our knowledge, this dataset will constitute the first large-scale collection of records for which early conditions in childhood can be linked with medium and long-term outcomes and a de-identified linked dataset containing the public data will be made available for research purposes. We will use these data to evaluate the impact of cash transfers in childhood on medium and long-term outcomes using identification techniques that rely on the construction of appropriate control groups. These include rejected applicants in the same county and year, ineligible mothers (e.g., divorced or abandoned mothers) in counties with strict eligibility requirements (e.g., widows only), and widowed mothers in states that did not establish, or were very late to establish, a mothers pension program (Aim 2). Finally, we are developing new econometric techniques for using imperfectly matched data. The current state of the art is to limit analysis to only perfectly or uniquely matched data, but this unnecessarily discards a great deal of information (all unmatched and multiply matched data) and ignores measurement error (some individuals are incorrectly matched). We are developing econometric techniques that will make use all of the information in the data (unique, multiple and no matches) by approximating the process that generates the data using reasonable parametric models and also estimating bounds (Aim 3).]